Specification · Bankability Index
Version BI-1.0 · published 14 August 2026 · effective from 14 August 2026 · supersedes nothing · next scheduled review 14 February 2027 · machine-readable record: bankability-index.json
Short form: Afrimintel Bankability Index BI-1.0 (2026). Any citation of a figure should state its depth tier (§4) and its state (§3).
The Index covers 159 rows across the African mining and critical-minerals assets carried in the Afrimintel deposits layer. Bankability is expressed differently by asset stage on purpose: forcing a single metric onto both a pre-FID study and a producing supermajor would be less honest, not more.
npv8_common is a first-order discount-rate restatement only, and must not be used to rank assets without price-normalisation first.Every published figure is exactly one of three states — Sourced named primary document with date, Derived published method with stated inputs, Absent gap named rather than filled. There is no fourth published state.
Pending and Inferred-propagated are verification working-tags, explicitly not part of the three-state model. They are surfaced during verification cycles rather than hidden. A working-tag is not a resting state: every entry is on a path to Sourced or Absent, carries a target verification date, and breaches the Correction Velocity SLA if it remains open past that date.
The register of open working-tags — with the reason each is open, its resolution path, and its target date — is published at sources-verification-log.json. Declared status as at 14 August 2026: 6 of 14 queued items are past their target verification date (reduced from 32 of 38 the same day, when credentialed ACLED Explorer sessions closed eighteen) — a Correction Velocity SLA breach under the rule stated above. It is declared here because an index that publishes an SLA and then quietly runs past it is worth less than one that says so. Any citation of this Index should treat working-tagged fields as unresolved and check the log for current status.
Where the same reserve field carries different figures on two Afrimintel surfaces, both are disclosed and neither is silently preferred; the adjudication is registered as an open item in the verification log against the operator's current reserve statement. This is enforced mechanically: contract-suite check-48 fails the build on any same-asset reserve divergence that is not disclosed on the surface carrying it. Four assets are currently in this state — Mogalakwena, Kansanshi, Sentinel, Obuasi.
Coverage and depth are published separately and must not be conflated. Breadth of coverage is not a claim of depth.
| Depth tier | Rows | What it carries |
|---|---|---|
| worked-DCF | 2 | Worked discounted-cash-flow reconciled to published anchors |
| sourced-study | 2 | Study economics read to primary technical report |
| sourced-headline | 9 | Headline study economics, named source, no rebuild |
| operating-markers | 72 | Sourced output, reserve life, cost-curve position |
| screened | 73 | Substrate only — screened, not worked |
| absent-stale | 1 | Prior economics withdrawn as superseded |
4 of 159 rows carry worked or sourced-study depth — 2 worked-DCF and 2 sourced-study read to the primary technical report. A further 9 carry sourced-headline economics: a headline figure with a named source, no rebuild. So 13 rows carry asset-level economics of any depth, of which only 4 are worked. The remaining 146 are substrate — 72 operating-marker rows, 73 screened, 1 withdrawn as superseded. Any citation of the Index must state which tier the figure comes from.
Stated at this granularity deliberately: collapsing 4 worked rows and 9 headline rows into a single "13 deep" figure would overstate depth by more than three times, which is precisely the defect this section exists to prevent.
Development / pre-FID — Sourced study economics: post-tax NPV, IRR and capex from the named technical report, with the study's discount rate and price deck stated so the reader sees the basis.
Producing — Sourced operating markers (output, reserve life) plus cost-curve position: C1 for copper, AISC/AIC for gold, cited to operator filings. Project NPV is marked Absent-operating: it is the wrong metric for a running mine.
Cash margin Derived — Sourced commodity price minus Sourced cost position. A transparent two-input derivation, not a modelled forecast.
Scenario operating margin Derived — Sourced annual volume × (Sourced scenario price − Sourced unit cost). Gross margin, 100% project basis, before royalties, tax and sustaining capital; not free cash flow, not attributable to any owner. Copper uses C1 (gross of sustaining capex, royalties, tax); gold uses AISC (closer to net).
npv8_common Derived — first-order restatement of issuer NPV to a common 8% rate, from issuer rate-sensitivity points where published, otherwise via the published level-annuity method. Where the level-annuity proxy is used for a pre-production asset it is flagged an indicative upper bound: the Doropo issuer rate-table shows the proxy overstates NPV8 by roughly 6% for such assets. Contested and data-layer-only assets are excluded from normalisation by design.
| Provision | Rule |
|---|---|
| Editorial authority | Nikesh Patel, Editor-in-Chief, Afrimintel Ltd — sole authority over the Index method. |
| Amendment | The method may be amended only by the Editor-in-Chief. Every amendment increments the version, is dated, is recorded in the change history, and is published here before it takes effect on any figure. |
| Material-change disclosure | A change that would alter a published figure's value or state is announced on the audit log at the time of restatement, with the prior value retained and dated, so a third party can see that the figure moved, when, what it was before, and on what basis. No advance-notice period is claimed at BI-1.0: a timed commitment that cannot be honoured under a sole-editor operating model would be worth less than an untimed one that is. An advance-notice window is a candidate for a later version once the working-tag queue is clear. |
| Non-material change | Clarifications that cannot alter a figure increment the patch component and are logged without a notice period. |
| Correction SLA | Material 7 business days; non-material 30 days, per the Quality Standard. |
| Independence | Afrimintel Ltd is independently owned — no strategic investors, sovereign holdings, or industry-corporate stakeholders. Independence Policy. |
| Conflicts | Operators and owners named are public-record parties to public assets; no commercial relationship is implied. |
| Version | Date | Change | Figures changed |
|---|---|---|---|
| BI-1.0 | 2026-08-14 | Initial formalisation. Documents and governs the method already in force (carried forward from Bankability Depth Methodology v1.3); assigns a version number and citation form; publishes the depth disclosure; states amendment and notice rules; publishes the working-tag register. | 0 |
This specification and governance record are published openly and may be cited. Redistribution of the underlying data layer is governed by the platform Terms. Licensing enquiries: nikesh@afrimintel.com.